Relationships between thermally induced residual stresses and architecture of epoxy-amine model networks

Abstract

The capability of epoxy-amine resins to develop residual stresses was studied as a function of temperature and network architecture. These residual stresses were induced while cooling epoxy-glass bilayers from temperatures higher than the network glass transition temperature, T g . This behavior was the result of the marked differences (α r - α g ), in linear thermal expansion coefficient of the two components, as evidenced by the measurement of α r for the epoxy networks under study. Various network architectures were selected, resulting from variation of (1) the chemical nature of both epoxide and curing agent, (2) the nature and relative amount of the chain-extensor agent, and (3) the stoichiometric ratio. Three ranges of cooling temperature were observed systematically: first, the range of temperatures above T g , where no stress has been detected, then an intermediate temperature range (from T g to T * ), where stresses develop quite slowly, and finally, the low temperature range (T < T * ), where a linear increase in stress accompanies the decrease of temperature. The two latter regimes were quantitatively characterized by the extent, T g -T * , of the first one and by the slope, SDR, of the second one. T g - T * values were shown to be governed by the T g of the network: the higher the T g , the larger the gap between T g and T * . This result was interpreted by accounting for the variation of relaxation rate at T g from one network to the other. It was also shown that a semiempirical relationship holds between SDR and T g : SDR decreases monotonically as T g increases. By inspecting the effects of network architecture in more details, it turned out that SDR is governed by the Young's moduli, E r (T- T g ), of the epoxy resins in the glassy state: the lower E r (T- T g ), the lower SDR in a series of homologous networks. As E r (T - T g ) values are known to be related to the characteristics of the secondary relaxation β, which depends, in turn, on crosslink density, SDR values were finally connected to the amplitude of the β relaxation processes. This finding was corroborated by the measurements on an antiplasticized dense network. Finally, data relative to thermoplastic-filled networks showed that the addition of thermoplastic reduces the development of residual stresses, whatever the system, is homogeneous or biphasic.